The Solution Structure of a Monocyclic Analog of Endothelin [1,15 Aba

Mar 1, 1994 - Extracting Information from the Temperature Gradients of Polypeptide NH Chemical Shifts. 1. The Importance of Conformational Averaging...
0 downloads 0 Views 1MB Size
686

J.Med. Chem. 1994,37, 656-664

The Solution Structure of a Monocyclic Analogue of Endothelin [ 1,15 Abal-ET-1, Determined by lH NMR Spectroscopy Murray Coles, Sharon L. A. Munro, and David J. Craik' School of Pharmaceutical Chemistry, Victorian College of Pharmacy, Monash University, 381 Royal Parade, Parkville, Victoria, Australia 3052 Received September 27, 1993.

The structure of [1,15 Abal-ET-1 has been determined in aqueous acetonitrile solution (10% acetonitrile, 1.5% acetic acid). [1,15 Abal-ET-1 is an analogue of endothelin (ET-1) in which the disulfide bridge linking residues 1and 15has been removed by replacement of the cysteine residues with the mimicking group a-aminobutyric acid (Aba). The structure has been determined by 1H NMR spectroscopy and simulated annealing calculations based on NOE constraints, 3&N-Ha scalar coupling constants, and amide-proton-exchange rates. Distance information was extracted from 2D NOESY spectra using full-relaxation matrix techniques (utilizing the program DISCON). The structure can be described in terms of three defined segments: a type I ,&turn over residues 5-8, a helix over residues 9-16, and a structured C-terminus over residues 17-21. The data indicate that some conformational averaging occurs throughout the peptide, particularly for residues 1-4 in the N-terminus, where no preferred conformation is present. The structure is compared with those previously reported for native ET-1. In general, removal of the disulfide bridge does not cause a major structural change in the helical and turn regions of the sequence, but increased structural disorder is noted a t the N-terminus. The implications of the monocyclic analogue's conformation for the pharmacological activity and the ETA/ETB selectivity of the endothelin family of peptides and analogues are described. The N-terminus is proposed to be a key structural region for differentiation of binding activity at the ETA and ETB receptor sites.

Introduction Endothelin (ET-1) is an endogenous peptide hormone which has attracted considerable attention due to its potent vasoconstrictor activity and extremelylong-lastingpressor effect. Originally isolated from cultured porcine aortic endothelial cells by Yanagisawa et al.,l it has since been found to be the first member of a family of closely related peptides, all members of which consist of 21 amino acid residues and are bicyclic with two disulfidebridges (linking residues 1-15 and 3-11). The sequences of ET-1 and two further endothelins, ET-2 and ET-3, are encoded in the human genome2and also in the genomes of other mammals. For example, ET-1 and ET-2 have been found in the dog314 and ET-3 in the rat6 and the rabbit.6 In addition, the family includes a fourth mammalian peptide, vasoactive intestinal contractor (VIC, also called endothelin-j3), identified in the genome of the mouse' and several snake toxins, including the earafotoxins(SRTX S6)isolated from the burrowing asp Atractaspis engaddensis.* Two distinct receptor subtypes have been identified for the endothelins, ETA^ and ETB.'O The activity of members of the endothelin family at these two receptors is different, with ET-1 and ET-2 being selective agonists at whereas all members of the endothelin family are approximately equipotent at ETB.'~ ETA is distributed predominantly in cardiovasculartissue and ETBin a wider range of tissues, includingthe kidneys and cerebe1lum.l3J4 The conservationof the disulfidebridges in all members of the endothelin family has led to investigations of their importance in the expression of biological activity. Monocyclic and linear analogues of ET-1, where the cysteine residues have been replaced by other residues, have been particularly useful for this purpose. Such peptides vary

* Author for correspondence.

Abstract published in Advance ACS Abstracts, January 15, 1994.

QQ22-2623/94/1837-0656$04.50/0

11

13

IS

17

19

21

ET-1

CHI CHI N H z -C

I

H

-COOH

Aha = a - amino butyric acid

11

13

1s

17

19

21

[1,15 AbapET-1

Figure 1. Amino acid sequences and disulfide connectivitiesof ET-1 and [1,15 Abal-ET-1. The molecular structure of the cysteine-replacingamino acid, a-aminobutyric acid (Aba),is also shown.

markedly in terms of their biological activity, receptor binding, and ETA/ETBselectivity.12Jb1e To date, however, the effect of such disulfide replacements on the solution conformation of the resulting peptides relative to that of native ET-1 is largely unknown. This report describes the solution structure of [1,15 Abal-ET-1, an analogue of ET-1 in which the cysteine residues have been replaced by a-aminobutyric acid, Aba (Figure 1). This amino acid is commonly used as an isosteric replacement for cysteine.19 The structure of this monocyclic ET-1 analogue has been determined in aqueous solution (10% acetonitrile, 1.5% acetic acid) by 2D 'H NMR spectroscopy and simulated annealing calculations. 0 1994 American Chemical Society

Journal of Medicinal Chemistry, 1994, Vol. 37, No. 5 657

Solution Structure of [1,15 Abal-ET-1

Table 1. Chemical Shifts (ppm) of [1,15 Abal-ET-1 at 293 K others residue NH Ha

m U

Abal

J G

3,

8.63

Cys3 Ser4

8.49 8.43

Leu6 Met7 hP8 Lys9

8.06 8.11 7.93 7.86 8.42

Ser5

0

I

1

I

4.0

3.0

I 2.0 F1

I

1.0

(pm)

Figure 2. TOCSY spectrum of [1,15Aba]-ET-l(600MHz, 298 K, 1.3 mM) showing connectivities between the NH protons and the side-chainspin systems. Residue numbers are indicated.Open squares indicate where cross peaks were observed but are not visible at this plot level.

@Q

12

3.93

Sera

GlulO Cysll Val12 TYrl3 Phel4 Abal5 His16 Leu17 Asp18

4.64 4.68 4.41 4.42

1.83 3.79

yCHs 0.94

3.14,2.93 3.91, 3.77 3.89,3.76

1.57 yCHz 1.68; 6CHs 0.99,0.93 2.09,l.W yCH2 2.53,2.42; cCH3 2.01 3.02,2.86 1.79 7CHz 1.40,1.36; SCHz 1.60; CCH~ 2.93;Nhz 7.46 8.38 4.05 2.10, 1.98 yCHz 2.42,2.32 7.88 4.50 3.05,3.03 7CH3 0.79,0.70 7.94 4.90 1.92 7.87 4.39 2.86,2.79 2,6H 6.87; 3,5H 6.66 7.87 4.43 3.09,2.92 2,6H 7.18; 3,5H 7.27; 4H 7.21 4.23 4.42 4.63 4.17

1.66

7CH3 0.84

1.63

7CHz 1.29,0.96; yCH3 0.51; GCHa 0.69 7CHz 1.32, 1.04; 7CH3 0.74;

Ile19

7.97 8.17 7.97 8.32 7.77

3.99 4.56 4.23 4.63 4.01

3.22,3.11 2H 8.70; 4H 7.20 1.56 7CHz 1.47; GCHs 0.80,0.76 2.86,2.71

Ile20

7.89

4.10

1.71

Trp2l

8.05

4.63

3.28,3.16

CHI 0.72 2H 71.6; 4H 7.56; 5H 7.05; 6H 7.13: 7H 7.29

293 K (mixing time = 200 ms). The chemical shifts of all proton resonances at this temperature are given in Table 1.

3

8.'6

8.4

8.'2 F2

8.0

7.'8

7.6

(ppml

Figure 3. Sequential connectivities observed for [1,15 AbalET-1 in the H,-HN fingerprint region of the 200-ms NOESY spectrumfor a 1.3mM solution at 293K. Intraresidue cross peaks are indicated by residue number.

Results Assignment of the NMR spectrum of [1,15 Abal-ET-1 was carried out using the sequential assignment method of Wuthrichm in which peaks are first assigned to an amino acid type and then their sequential position is assigned by determining NOESY connectivities between adjacent residues. For [1,15AbaI-ET-l, the side-chainspin system of each amino acid residue was identified mainly from amide-proton connectivities in the 2D TOCSY spectrum in HzO (Figure 2). Ambiguities were resolved by reference to the DQF-COSY spectrum. Early in the process, the side chains of the three serine residues (Ser2, Ser4, and Ser5) were readily identified, and examination of the NOESY spectrum led to the assignment of a series of daN(i, i + 1)connectivities spanning residues 1-6. Identification of the unique side-chain spin systems of Lys9 and Val12 similarly led to the assignment of residues 8-13, as GlulO was distinguished from the similar spin system of Met7, and Tyrl3 from Phel4. Connectivities in 2D NOESY spectra between aromatic and H resonances in His16 and Trp2l identified these residues, leading to the assignment of residues 15-21, thus completing the assignments. The back-bone sequential assignments are summarized in Figure 3, which shows the connectivities observed for [1,15 Abal-ET-1 in the NOESY spectrum at

Stereospecific assignments were made for the prochiral H resonances of 5 of the 11AMX spin systems. This was completed from considerations of cross-peak patterns in the 2D TOCSY spectra in deuterated solvent in conjunction with peak intensities in NOESY experimentseZ1In residues where stereospecific assignment was obtained, the x1torsion angle of the predominant rotamer was also determined. In cases where stereospecific assignment could not be obtained, either the prochiral Hfl resonances were very similar in chemical shift or the data could not be interpreted unambiguously, presumably due to rotation about the x1torsion angle. Stereospecific assignment was also possible for the y-methylene protons of Ile19 and Ile20, and in these cases, the predominant rotamer about x1 or x2 was also determined. A summary of sequential and long-range NOE connectivities is presented in Figure 4. Also indicated are 'JHN-H~ scalar coupling constants measured as greater than 8.0 Hz or less than 6.0 Hz and amide protons identified as exchanging slowly with solvent. A qualitative analysis of the data provides evidence of elements of secondary structure in several regions of the peptide, as described below. A number of daN(i, i + 31, daN(i, i + 41, and dap(i, i i3) connectivities are observed for residues 9-16 and are typical of a helical structure. Strong d"(i, i + 1) connectivities, also indicative of helix, were observed in this region. However, with the exception of LysQ HNGlulO HN and GlulO HN-Cysll HN, extensive shift coincidenceof the amide-proton resonances involved made quantitation of these cross peaks difficult due to their close proximity to the diagonal. Further support for a helical structure comes from the series of slowly exchanging amide protons from residues 12-16, suggesting the formation of intramolecular hydrogen bonds. A strong d"(i, i + 1)cross peak between residues Leu6 and Met7, together with a number of weak connectivities between residues Ser5 and Asp8 and 3 J ~ coupling ~-~a

Coles et al.

658 Journal of Medicinal Chemistry, 1994, Vol. 37,No. 5 1

5

15

10

X S CS S L

0.

0

HN Ex.

VA

V

--M

a

0.

0.

V

A A

M

M

0th"

20

M D K Q C V Y F X H L DI I W

a

&a

&

M M 0

Figure 4. Summary of the observed NOE connectivities for [1,15Abal-ET-1in the 200-ms NOESY spectrum at 293 K. The observed NOEs are classified into four levels as quantified by the height of the bars. Hatched bars and dashed lines indicate where quantification of the intensities was not possible due to spectral overlap. Medium and long connectivitiesare also marked. Filled and open circles representconnectivitiesbetween backbone and side chains,respectively. The number of occurrencesof each scalar . ~coupling constants type is indicated above the bar. 3 J ~ measured as greater than 8.0 Hz (A)or less than 6.0 Hz (V)are indicated, as are amide protons identified as exchanging slowly with solvent (0). constants for Leu6 and Met7 of 8.0 Hz, respectively,is indicative of a turn structure about residues 5-8. This is supported by the observation of a slowly exchanging amide proton of Asp8. The large number of connectivities observed in the C-terminal region involving residues 17-21, especiallyd m (i, i l),and numerous d(i, i 2) cross peaks indicate that this region of the peptide is also structured. The precise nature of the structure, however, is not apparent from a qualitative analysis of the NOEs. Over the complete peptide sequence, only three connectivities have a span of more than four residues: two involve Ser5 and Val12 and another involves the cysteine residues 3 and 11,which form the disulfide bridge. These long-range NOESassist in defining the spatial relationships between the local structured elements noted above. Given the relatively small size of the peptide, the existence of any long-range NOEs provides good support for the presence of a well-defined global fold. The qualitative analysis of NOE and exchange data described above strongly suggests that [1,15 Abal-ET-1 adopts a well-defined conformation in solution. To more precisely determine the nature of this conformation, a series of simulated annealing structure calculations was performed using the experimental NOE data as distance constraints.

+

+

Structure Calculations Quantitative analysis of the NOE intensities yielded 144 sequential and 31 medium- and long-range distance constraints from experiments a t 293 K and an additional

Figure 5. Stereoview of the lowest-energy structure for [1,15 Abal-ET-1.Back-bone atoms are shown for all residues, as are heavy atoms in the disulfide bridge. Side chains are not shown for clarity. three sequential and two medium-range constraints from experiments at 288 K. Six torsion angle constraints were included on the basis of the observations given in Figure 4. The @I angles of Leu6, Lys9, and Abal5 were constrained to -60' f 30°, and those of Met7, Ile19, and Ile20 were constrained to -120' f 50'. Three further XIor xztorsions were included for non-AMX residues which had been stereospecifically assigned (Val12 XI = 60°, Ile19,Ile20 xz = -60'). The XI angles of AMX residues were not constrained in the simulation. The structure calculation protocol outlined in the Experimental Section yielded sets of 40 structures for [l,15 Abal-ET-1. Preliminary sets of structures were calculated without H-bond constraints in order to determine possible H-bond acceptors for the slowly exchangingamide protons indicated in Figure 4. The average distance from each slowly exchanging amide proton to possible H-bond acceptors was measured to determine the likely H-bond acceptor. Three hydrogen bonds are suggested, namely Asp8 HN Ser5 CO, Tyrl3 HN Lys9 CO, and His16 HN Val12 CO, and these were included as constraints in the final structure calculations. No acceptor was unambiguouslysuggested by this analysis for the two slowly exchanging amide protons in the C-terminal region of the peptide, Ile19 HN and Ile20 HN. The final set of 40 structures was calculated after the inclusion of dihedral and H-bond constraints. Ten of these structures were rejected on the basis of violation of the experimental constraints, leaving a set of 30 accepted structures. The root mean square deviation of the structures on the basis of back-bone atoms, calculated as the average of all possible pairwise comparisons, was 3.2 A. A stereoview of the back-bone atoms of the lowestenergy structure calculated for [1,15 Abal-ET-1 is shown in Figure 5. Although therms deviation of the global superimposition is high, the level of agreement obtained between structures is demonstrated if the peptide is considered as consisting of several segments. Two recognizable elements of sec-

--

-

Solution Structure of [1,15 Abal-ET-1

Journal of Medicinal Chemistry, 1994, Vol. 37, No. 5 659

Table 2. Average Back-Bone Torsion Anglesn 4 and $ for [1,15 Abal-ET-1 residue no. 1 2 3 4 5 6 7 8 9 10 11

4

$

-72 -74 -60 -174 -57 (-60)b -81 (-90) -126 -131 -76 -121

153 123 172 25 172 -17 (-30) -14 (0) 171 -45 14 -1

residue no.

4

$

12 13 14 15 16 17 18 19 20 21

-126 -72 -88 -102 -110 -135 77 -143 -107 -46

-45 -10 -22 -35 66 33 94 105 34

a Angles in degrees. b Values in parentheses are standard torsion angles for a type 18-turn.23

118 19

1

2

3

4

5

5

7

8

9

10 11 12 13 14 15

s

118 19

16 17 18 19 20 21

Residue No.

Figure 6. Angular order parametersfor back-bonetorsion angles of the 30 structures calculated for [1,15 Abal-ET-1. For each residue, 4 and (I.are shown as filled and hatched bars, respectively.

ondary structure have been identified, namely a &turn at residues 5-8 and a helix for residues 9-16. In addition, a well-defined region also exists for the C-terminus (residues 17-21). The calculated structure is well defined within these individual segments;however,the relationship of the segments to each other varies considerably, leading to poor superimposition of the molecule as a whole. A fourth segment comprising the N-terminus (residues 1-4) does not show a preferred conformation. The average back-bone torsion angles for the calculated structures are shown in Table 2. Angular order parametersz2were calculated to provide ameasure of the variation of individual torsion angles between structures. Values of this parameter approach unity where the angles are in complete agreement, falling to zero where they are randomly distributed. Figure 6 shows angular order parameters for the angles of each residue for the set of 30 calculated structures. Values of this parameter are generally high within the segments mentioned above. Slightly lower values for angles at the borders of the regions indicate a higher degree of local variation, suggesting these residues act as hinges for segmental motion. The disordered nature of the N-terminal segment (residues 1-4) is also clearly demonstrated. Given the nature of the structure, further analysis was done separately on the individual segments. These are referred to as the @-turn,helix, C-terminus, and N-terminus segments. The nature of the structure calculated in each segment is discussed in detail in the following sections. &Turn. The structure of the segment comprising residues 5-8 can be described as a type I 8-turn stabilized

Figure 7. Stereoviews of 30 calculated structuresfor [1,15AbalET-1superimposed on the basis of back-bone atoms: (a) &turn region from residues 5 to 8, (b) helical region from residues 9 to 16, and (c) C-terminalregion from residues 17 to 21. Also shown in (b)but not included in the superimpositionare the C-terminal region residues 17-21. The range of conformational space occupied by the C-terminal region relative to the helix is demonstrated. A good superimposition(rmsd = 0.42 A) is obtained within the C-terminal region, as shown in (c). by a hydrogen bond between Asp8 HN and Ser5 CO. In most structures, an additional hydrogen bond is formed between Met7 HN and Ser5 CO. The involvement of Asp8 HN in a hydrogen bond is supported by its slow exchange with DzO solvent; however, as both Met7 HN and Met7 H a are unresolved, the involvement of Met7 HN in a hydrogen bond could not be confirmed. Standard torsion angles for a type I @-turnz3are shown in Table 2 for comparison with the derived angles. For the central residues (i.e., 6 and 7) of the proposed &turn, there is a close match of the experimentally determined angles to those expected for a standard type I turn. The set of 30 accepted structures shows no deviations greater than 0.3 A from the NOE constraints over this segment and no deviations from the dihedral constraints. A superimposition of this set on the basis of back-bone atoms is shown in Figure 7a. The rmsd of this superimposition, calculated as the average of all possible pairwise comparisons, is 0.60

A.

Helix. The pairwise rms deviation of the superimposition over the helical region from residues 9 to 16 is 0.52 A. Analysis of the calculated structures for the deviation of distance and dihedral constraints revealed that several constraints, mostly of the type daN(i, i + 1)and d&, i + 11, are violated, albeit to a small degree (